ebook img

Anomalous statistical properties of the critical current distribution in superconductor containing fractal clusters of a normal phase PDF

11 Pages·0.27 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Anomalous statistical properties of the critical current distribution in superconductor containing fractal clusters of a normal phase

Anomalous statistical properties of the critical current distribution in superconductor containing fractal clusters of a normal phase Yuriy I. Kuzmin Ioffe Physical Technical Institute of the Russian Academy of Sciences, Polytechnicheskaya 26 St., Saint Petersburg 194021 Russia, 3 and State Electrotechnical University of Saint Petersburg, 0 0 Professor Popov 5 St., Saint Petersburg 197376 Russia 2 e-mail: [email protected]ffe.ru; [email protected] tel.: +7 812 2479902; fax: +7 812 2471017 n (February 2, 2008) a J 4 1 Statisticalpropertiesofthecriticalcurrentdistributioninsuperconductorwithfractalclustersof ] anormalphaseareconsidered. Itisfoundthatthereistherangeoffractaldimensionsinwhichthe n varianceandexpectationforthisdistributionincreasesinfinitely. Simpletechniqueofavoidingsuch o adivergencebytheuseoftruncateddistributionsisproposed. Itissuggestedthatthemostcurrent- c carrying capability of a superconductor can be achieved by modifying the cluster area distribution - r in such a way that theregime of giant variance of critical currents will berealized. p u s . I. INTRODUCTION t a m Considerable recent attention is being drawn to the fractal behavior of magnetic flux in type-II superconductors - [1]- [3]. High temperature superconductors(HTS) containing clusters ofcorrelateddefects [4,5]are of special interest d in this field. The case of clusters with fractal boundaries provides new possibilities for increasing the critical current n value [2,6]. By virtue of the capability to trap a magnetic flux, such clusters can appreciably modify magnetic and o transportproperties ofsuperconductors[3,7,8]. The distributionofthe criticalcurrentsinsuperconductorcontaining c [ the normal phase clusters with fractal boundaries has unusual statistical properties, and it is these features that will interest us. 1 Let us consider a superconductor containing columnar inclusions of a normal phase, which are out of contact with v one another. These inclusions may be formed by the fragments of different chemical compositions, as well as by 3 the domains of the reduced superconducting order parameter. The similar columnar defects can readily be created 3 during the film growth process [5,9,10]. In the course of the cooling below the critical temperature in the magnetic 2 1 field (“field-cooling“ regime) the magnetic flux will be trapped in the isolated clusters of a normal phase so the two– 0 dimensional distribution of the flux will be created in such a superconducting structure. When the transport current 3 is passed transversely to the magnetic field, this one is added to all the persistent currents, which circulate around 0 the normal phase clusters and keep the trapped magnetic flux to be unchanged. By the cluster we mean a set of the / columnar defects, which are united by the common trapped flux and are surrounded by the superconducting phase. t a Inasmuch as the distribution of the trapped magnetic flux is two–dimensional, instead of dealing with an extended m object, which indeed the normal phase cluster is, we will consider its cross-section by the plane carrying a transport - current. As was first found in Ref. [2], clusters of a normal phase can have fractal boundaries, and this feature has a d significanteffect on the dynamics of the trapped magnetic flux [3,8]. In the subsequent considerationwe will suppose n that the characteristic sizes of the normal phase clusters far exceed both the coherence length and the penetration o depth. ThisassumptionagreeswellwiththedataontheclusterstructureinYBCOfilms[2,3,10],aswellaswillallow c : us to highlight the role played by the cluster boundary in the magnetic flux trapping. v i X II. GIANT DISPERSION OF CRITICAL CURRENTS r a Supposethatthereisasuperconductingpercolationclusterintheplaneofthefilmwhereatransportcurrentflows. Such a structure provides for an effective pinning, because the magnetic flux is locked in finite clusters of a normal phase. Whenthe transportcurrentisincreased,the trappedmagneticfluxremainsunchangeduntilthe vorticesstart to break away from the clusters of pinning force weaker than the Lorentz force created by the current. As this takes place, the vortices must cross the surrounding superconducting space, and they will first do that through the weak links,whichconnectthenormalphaseclustersbetweenthemselves. SuchweaklinksformeasilyinHTScharacterized byanextremelyshortcoherencelength. Diversestructuraldefects,whichwouldsimplycauseanadditionalscattering atlongcoherencelength,giverisetotheweaklinksinHTS.Weaklinksarisereadilyontwinboundaries,andmagnetic 1 fluxcaneasilymovealongthem[11]. Whateverthemicroscopicnatureofweaklinkscouldbe,theyformthechannels forvortextransport. Accordinglytoweaklinkconfigurationeachnormalphaseclusterhasitsownvalueofthecritical current, which contributes to the total distribution. By the critical current of the cluster we mean the current of depinning, that is to say, such a current at which the magnetic flux ceases to be held inside the cluster of a normal phase. The critical current distribution is related to the cluster area distribution, because the cluster of a larger size has more weak links over its boundary with the surrounding superconducting space, and thus the smaller current of depinning [3]. In the practically important case of YBCO films with columnar defects the exponential distribution of the cluster areas can be realized [2], which is the special case of gamma distribution. The exponential distribution has only one characteristicparameter(meanclustersize),sotherearenotmanypossibilitiestomodifythegeometricmorphological properties of the clusters in this simplest case. By contrast, in the case of gamma distribution there is an additional wayforoptimizingtheclusterstructureofthecompositesuperconductorbythecontroloftwoindependentparameters in the course of the film growth. One of the aims of the present work is to find how the cluster area distribution shouldbeoptimizedinordertogetthehighestcurrent-carryingcapabilityofasuperconductor. InRef. [7]thecritical current distribution was derived in the case of gamma distribution of fractal clusters areas, which has the following probability density Agexp(−A/A ) 0 w(A)= (1) Γ(g+1)Ag+1 0 where Γ(ν) is Euler gamma function, A is the cluster area, A > 0 and g > −1 are the parameters of gamma 0 distribution that control the mean area of the cluster A = (g+1)A and its variance σ2 = (g+1)A2. For further 0 A 0 consideration it is convenient to introduce the dimensionless area of the cluster a ≡ A/A, for which the distribution function of Eq. (1) can be rewritten as: (g+1)g+1 w(a)= agexp(−(g+1)a) (2) Γ(g+1) Themeandimensionlessareaoftheclusterisequaltounity,whereasthevarianceisdeterminedbyoneparameteronly: σ2 =1/(g+1). The probability density of Eq. (2) is presented in Fig. 1 for the characteristic values of g-parameter. a The critical current distribution has the following form [7] 2Gg+1 f(i)= i−(2/D)(g+1)−1exp −Gi−2/D (3) DΓ(g+1) (cid:16) (cid:17) where 2 θθ D G ≡ θg+1−(D/2)exp(θ)Γ(g+1,θ) (cid:18) (cid:19) D θ ≡ g+1+ 2 Γ(ν,z) is the complementary incomplete gamma function, i ≡ I/I is the dimensionless electric current, I = c c α(A G)−D/2 is this the criticalcurrentofthe resistivetransition,αisthe formfactor,andD is thefractaldimension 0 of the cluster perimeter. The value of D specifies the scaling relation P1/D ∝A1/2 between perimeter P and area A of the cluster [12,13]. TheprobabilitydensitycurveforcriticalcurrentdistributionofEq.(3)hastheskewbell-shapedformwithinherent broad“tail“extendedovertheregionofhighcurrents(seeFig.2). Asmaybeseenfromthisgraph,thecriticalcurrent distribution spreads out with some shift to the right as g-parameterdecreases. This broadening can be characterized by the standard deviation of critical currents 2 D Γ(g+1−D) Γ(g+1−D/2) σi =G2 − (4) s Γ(g+1) Γ(g+1) (cid:18) (cid:19) The standard deviation grows nonlinearly with increase in the fractal dimension, as is illustrated in Fig. 3. The peculiarity of the distribution of Eq. (3) is that its variance becomes infinite in the range of fractal dimensions D ≥ g+1. The distributions with divergent variance are known in probability theory - the classic example of that 2 kindisCauchydistribution[14]. However,suchananomalousfeatureofexponential-hyperbolicdistributionofEq.(3) isofspecialinterest,inasmuchasthecurrent-carryingcapabilityofasuperconductorwouldbeexpectedtoincreasein theregionofgiantvariance. Thenthestatisticaldistributionofcriticalcurrentshasaveryelongated“tail“containing the contributions from the clusters of the highest depinning currents. The distribution of Eq. (3) has one even more striking feature: its mathematical expectation, which represents the mean critical current Γ(g+1−D/2) D i= G2 (5) Γ(g+1) is also divergent in the range of fractal dimensions D > 2(g+1). At the same time, the mode of the distribution modef(i)=(G/θ)D/2 remains finite for all possible values of the fractal dimension 16D 62. The reason for divergence of the mean critical current consists in the behavior of the cluster area distribution of Eq.(2). AsmaybeseenfromFig.1,thegraphofthedistributionfunctionofEq.(2)takesessentiallydifferentshapes depending on the sign of g-parameter- from the skew unimodal curve (1) (at g >0) to the monotonic curve (3) with hyperbolic singularity at zero point (at g <0). In the borderline case of g =0, which separates these different kinds of the functions, the distribution obeys an exponential law (curve (2)). It is just for negative values of g-parameter that the mean critical current diverges. The contribution from the clusters of small area to the overall distribution grows at g < 0. Since the clusters of small size have the least number of weak links over a perimeter, they can best trapthe magnetic flux. Therefore,anincrease of the part of smallclusters in the area distribution ofEq. (2) leads to a growth of the contribution with high depinning currents made by these clusters in the critical current distribution of Eq. (3). Just as a result of this feature the mean critical current diverges at g < 0. Nevertheless, the total area between the curve of the probability density (like curve (3) in Fig. 2) and the abscissa remains finite by virtue of the normalization requirement. III. TRUNCATED DISTRIBUTION OF CRITICAL CURRENTS Obviously, the proper critical current cannot be infinitely high as well as the clusters of infinitesimal area do not really exist. There is the minimum value of the normal phase cluster area A , which is limited by the processes of m thefilmgrowth. SoinYBCObasedcomposites,whichwerepreparedbymagnetronsputteringonsapphiresubstrates with aceriumoxidebuffer sublayer[3],the samplevalue ofminimum areaofthe normalphase cluster hasbeen equal to A = 2070 nm2 at mean cluster area A = 76500 nm2, that corresponds to the lower bound of the dimensionless m area of the cluster a ≡ A /A = 0.027. In view of this limitation, we will describe the distribution of the normal m m phase cluster areas by the truncated version of the probability density of Eq. (2): h(a−a ) w(a|a>a )= m w(a) (6) m 1−W (a ) m 1 for x>0 where γ(ν,z) is the complementary gamma function, h(x)≡ is the Heaviside step function, 0 for x<0 (cid:26) am γ(g+1,(g+1)a ) m W (a )≡ w(a)da= (7) m Γ(g+1) Z 0 is the truncation degree, which is equal to the probability Pr{∀a<a } to find the cluster of area smaller than the m least possible value a in the initial population. m The expression of Eq. (6) gives the conditional distribution of probability, for which all the events of finding a cluster of area less than a are excluded. Thus the truncation provides a natural way to fulfil our initial assumption m that the cluster size has to be greater than the coherence and penetration lengths. New distribution of cluster areas gives rise to the truncated distribution of the critical currents: h(i −i) f(i|i6i )= m f(i) (8) m 1−W (a ) m where i ≡(G/(g+1)a )D/2 is the upper bound of the depinning current, which corresponds to the cluster of the m m least possible area a . m Then, instead of Eqs. (4) and (5), the standard deviation and mean critical current are 3 2 ∗ D Γ(g+1−D,(g+1)am) Γ(g+1−D/2,(g+1)am) σ = G2 − i s Γ(g+1,(g+1)am) (cid:18) Γ(g+1,(g+1)am) (cid:19) ∗ Γ(g+1−D/2,(g+1)am) D i = G2 Γ(g+1,(g+1)a ) m For the truncated distribution of Eq. (8) the possible values of depinning currents are bounded from above by the quantity i , therefore the mean critical current as well as the variance do not diverge any more. Both of m these characteristics are finite for any fractal dimensions, including the case of maximum fractality (D = 2). The corresponding graphs for standard deviation are presented in Fig. 4. All the curves are calculated at g = −0.2 (as for the main curve (2) of the mean critical current in Fig. 5). In this case the variance for initial distribution of the criticalcurrents is infinite, so no graphfor W (a )=0 is shownin Fig. 4 at all. The dependence of the mean critical m current on the fractal dimension in the case of truncated distribution is demonstrated in Fig. 6. The corresponding graphs in Figs. 4 and 6 are drawn for the same values of truncation degree. The truncation degree is related to the least possible area of the cluster by the equation (7). The values of W(a ) and a involved in Figs. 4 and 6 are m m presented in the Table I. The degree of truncation gives the probability measure of the number of normal phase clusters that have the area smallerthan the lowerbound a in the initialdistribution. Ifthe magnitude ofW (a ) is sufficiently small (no more m m than several percent), then the procedure of truncation scarcely affects the initial shape of the distribution and still enablesthecontributionfromtheclustersofzeroarea(therefore,ofinfinitelyhighdepinningcurrent)tobeeliminated. It is interesting to note that the very similar situation occurs in analyzing the statistics of the areas of fractal islands in the ocean [12]. The island areas obey the Pareto distribution, which also has the hyperbolic singularity at zero point that causes certain of its moments to diverge. For exponential distribution of the cluster areas, which is valid in the above-mentioned case of YBCO films [3], the probability to find the cluster of area smaller than a in the m sampling is equal to Pr{∀a<a }=2.7% only. In principle, the truncation procedure could be made here, too, but m there isno needforthat,because the contributionofinfinitesimally smallclustersis finite atg =0 (andequaltozero at g >0). Referring to Figs. 4 and 6 it can be seen that the dependencies of the standard deviation and the mean critical current on the fractal dimension become smoother and smoother with increase in the degree of truncation. At the sametime,theinherenttendency ofinitialdistributionisstillretained: asthe fractaldimensionincreases,the critical current distribution broadens out (the variance grows,see Fig. 4), moving towards higher currents (the mean critical current grows, see Fig. 6). As may be seen from Fig. 2, this trend is further enhanced with decreasing g-parameter. The most current-carrying capability of a superconductor should be achieved when the clusters of small size, which havethehighestcurrentsofdepinning,contributemaximallytotheoveralldistributionofthecriticalcurrents. Sucha situationtakesplacejustintheregionofgiantvarianceofcriticalcurrents. Sofar,theleastmagnitudeofg-parameter (equaltozero)hasbeenrealizedinYBCOcompositescontainingnormalphaseclustersoffractaldimensionD =1.44 [2]. The critical currents of superconducting films with such clusters are higher than usual [6,9,10]. It would thus be expected to further improve the current-carrying capability in superconductors containing normal phase clusters, which will be characterized by area distribution with negative magnitudes of g-parameter, especially at the most values of fractal dimensions. Thus, it has been revealed that the distribution of the critical currents in superconductor with fractal clusters of a normal phase has the anomalous statistical properties, implying that its variance and expectation diverges in the certainrangeoffractaldimensions. Itmaybe expectedthatthe mostcurrent-carryingcapabilityofasuperconductor can be achieved by optimization of the cluster area distribution that involves reducing g-parameter with concurrent increasing the fractal dimension. IV. ACKNOWLEDGEMENTS This work is supported by the Russian Foundation for Basic Researches (Grant No 02-02-17667). [1] R.Surdeanu,R.J.Wijngaarden,B.Dam,J.Rector,R.Griessen, C.Rossel, Z.F.Ren,andJ.H.Wang,Phys.Rev.B58, 12467 (1998); C. J. Olson, C. Reichhardt,and F. Nori, Phys.Rev. Lett.80, 2197 (1998). 4 [2] Yu.I. Kuzmin, Phys.Lett. A 267, 66 (2000). [3] Yu.I. Kuzmin, Phys.Rev.B 64, 094519 (2001). [4] M.Baziljevich,A.V.Bobyl,H.Bratsberg,R.Deltour,M.E.Gaevski,Yu.M.Galperin,V.Gasumyants,T.H.Johansen,I. A.Khrebtov,V.N.Leonov,D.V.Shantsev,andR.A.Suris,J.Phys.(Paris)IV6,C3-259(1996);M.Prester,Supercond. Sci. Technol. 11, 333 (1998); M. Prester, Phys. Rev. B 60, 3100 (1999); B. Dam, J. M. Huijbregtse, and J. H. Rector, Phys.Rev.B 65, 064528 (2002). [5] E. Mezzetti, R. Gerbaldo, G. Ghigo, L. Gozzelino, B. Minetti, C. Camerlingo, A. Monaco, G. Cuttone, and A. Rovelli, Phys.Rev.B 60, 7623 (1999). [6] Yu.I. Kuzmin, Tech.Phys. Lett. 26, 791 (2000). [7] Yu.I. Kuzmin, Phys.Lett. A 281, 39 (2001). [8] Yu.I. Kuzmin, Phys.Solid State 43, 1199 (2001). [9] Yu.I. Kuzmin and I.V. Plechakov, Tech.Phys. Lett. 25, 475 (1999). [10] Yu.I. Kuzmin, I.V. Pleshakov, and S. V.Razumov, Phys.Solid State 41, 1594 (1999). [11] C.A.Duran,P.L.Gammel, R.Wolfe,V.J.Fratello, D.J.Bishop, J.P.Rice,andD.M.Ginsberg, Nature(London)357, 474 (1992); C. A. Duran, P. L. Gammel, D. J. Bishop, J. P. Rice, and D. M. Ginsberg, Phys. Rev.Lett. 74, 3712 (1995); U. Welp, T. Gardiner, D. O. Gunter, B. W. Veal, G. W. Crabtree, V. K. Vlasko-Vlasov, and V. I. Nikitenko, Phys. Rev. Lett.74, 3713 (1995); R. J. Wijngaarden, R. Griessen, J. Fendrich,and W.-K.Kwok, Phys. Rev.B 55, 3268 (1997). [12] B. B. Mandelbrot, The Fractal Geometry of Nature (Freeman, San Francisco, 1982). [13] B. B. Mandelbrot, Fractals: Form, Chance, and Dimension(Freeman, San Francisco, 1977). [14] D.Hudson,Statistics (CERN, Geneva,1964). TABLEI. Degree of truncation of the cluster area distribution and corresponding least possible area of the normal phase cluster at g = - 0.2 Truncation degree W(am) Minimum area am Markers in Figs. 4, 6 0.01% 1.144×10−5 a 0.1% 2.034×10−4 b 1% 3.622×10−3 c 5 2.0(cid:13) 1.5(cid:13) 3(cid:13) w(cid:13) 1.0(cid:13) 1(cid:13) a(cid:13) 0.5(cid:13) 2(cid:13) 0.0(cid:13) 0.5(cid:13) 1.0(cid:13) 1.5(cid:13) 2.0(cid:13) a(cid:13) FIG.1. The distribution of the areas of normal phase clusters at different values of g-parameter. Curve (1) corresponds to thecase of g=1; curve(2) of g=0; curve(3) of g=−0.5. The arrow indicates themean cluster area. 6 0.5(cid:13) 1(cid:13) 0.4(cid:13) 0.3(cid:13) 2(cid:13) f(cid:13) 0.2(cid:13) 3(cid:13) 0.1(cid:13) 0(cid:13) 1(cid:13) 2(cid:13) 3(cid:13) 4(cid:13) 5(cid:13) i(cid:13) FIG.2. The critical current distribution for the fractal dimension of the cluster boundary D =1.5. Curve (1) corresponds to thecase of g=1; curve(2) of g=0; curve(3) of g=−0.5. 7 30(cid:13) 25(cid:13) 20(cid:13) 1(cid:13) 2(cid:13) 3(cid:13) 4(cid:13) (cid:13)i 15(cid:13) s(cid:13) (cid:13) 10(cid:13) 5(cid:13) 0(cid:13) 1.0(cid:13) 1.2(cid:13) 1.4(cid:13) 1.6(cid:13) 1.8(cid:13) 2.0(cid:13) D(cid:13) FIG.3. Influence of the fractal dimension of the cluster boundary on the standard deviation of critical currents. Curve (1) corresponds to the case of g=0.25; curve(2) of g=0.5; curve(3) of g=0.75; curve(4) of g=1. 8 120(cid:13) 90(cid:13) i(cid:13) *(cid:13) 60(cid:13) s(cid:13) (cid:13) a(cid:13) b(cid:13) c(cid:13) 30(cid:13) 0(cid:13) 1.0(cid:13) 1.2(cid:13) 1.4(cid:13) 1.6(cid:13) 1.8(cid:13) 2.0(cid:13) D(cid:13) FIG.4. Standarddeviationgraphsfortruncateddistributionofthecriticalcurrentsatg=−0.2withthedifferentdegreeof truncation: curve(a) is drawn for W(am)=0.01%; curve(b) for W(am)=0.1%; curve(c) for W(am)=1%. 9 40(cid:13) 30(cid:13) 1(cid:13) 2(cid:13) (cid:13)i 20(cid:13) 3(cid:13) 10(cid:13) 4(cid:13) 0(cid:13) 1.0(cid:13) 1.2(cid:13) 1.4(cid:13) 1.6(cid:13) 1.8(cid:13) 2.0(cid:13) D(cid:13) FIG.5. Influence of the fractal dimension of the cluster boundary on the mean critical current. Curve (1) corresponds to thecase of g=−0.4; curve(2) of g=−0.2; curve(3) of g=0; curve(4) of g=1. 10

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.